基于BP神经网络的深部巷道围岩力学参数反分析  

Back Analysis of Mechanical Parameters of Surrounding Rock in Deep Roadway Based on BP Neural Network

作　　者：李坤铎 LI Kunduo(School of Civil Engin.,Architecture and Environment,Hubei Univ.of Tech.,Wuhan 430068,China)

机构地区：[1]湖北工业大学土木建筑与环境学院,湖北武汉430068

出　　处：《湖北工业大学学报》2024年第2期116-120,共5页Journal of Hubei University of Technology

摘　　要：基于BP神经网络算法原理,借助matlabR2021b神经网络工具箱建立深部巷道围岩力学参数位移反分析模型,利用正交试验和Flac 3D数值模拟软件建立神经网络的学习训练样本,对深部巷道的四个围岩力学参数粘聚力C、内摩擦角φ、泊松比ν、弹性模量E进行反演计算。结果表明:将参数反演结果代入Flac 3D有限元数值模拟软件,计算出的巷道拱顶沉降和两帮收敛值与实际监测值相比非常接近,相对误差小、精度高。通过这种方法获取的围岩力学参数是有价值的,可以较为精确地获取深部巷道的围岩力学参数,从而为深部巷道的稳定性分析及巷道支护设计提供科学依据。Based on the principle of BP neural network algorithm,the back analysis model of mechanical parameter displacement of surrounding rock of deep roadway is established with the help of MATLAB r2021b neural network toolbox.The learning samples of neural network are established by using orthogonal test and FLAC3D numerical simulation software.The cohesion,internal friction angle,Poisson's ratio and elastic modulus of four mechanical parameters of deep tunnel surrounding rock are calculated in reverse direction.The results show that by substituting the parameter inversion results into the FLAC3D finite element numerical simulation software,the calculated roadway vault settlement and two side convergence values are very close to the actual monitoring values,with small relative error and high precision.The mechanical parameters of sedimentary rocks obtained by this method are useful.More precise rock mechanical parameters of the deep roadway can be obtained,so as to provide a scientific basis for the overall layout of deep roadway and roadway design support.

关 键 词：BP神经网络 FLAC 3D数值模拟 巷道施工 

分 类 号：TQ320.6[化学工程—合成树脂塑料工业]